Skip to content
Chalk Labs / AI SEO
Definition

What Is a Knowledge Graph?

By Rahil Jain, founder of Chalk LabsUpdated How we research

The short answer

A knowledge graph is a database of real-world things, called entities, and the relationships between them. Google's Knowledge Graph powers knowledge panels and helps it tell apart things with the same name. You cannot ask Google to create a panel for you. You can make your facts clear, consistent and verifiable so its systems can recognise you.

  • A knowledge graph links entities (people, places, businesses, ideas) through stated relationships, not just words 1.
  • Google said in 2020 its Knowledge Graph held over 500 billion facts about five billion entities 2.
  • Google says it does not manually create or delete knowledge panels, but subjects can verify and suggest changes 3.
  • Clear Organization markup, consistent profiles and public references such as Wikidata give systems more to match 45.

What is a knowledge graph?

A knowledge graph is a way of storing information as things and the links between them. A thing (called an entity) is a person, place, company, product or idea. A link (called a relationship) says how two things connect, such as 'founded', 'located in' or 'works at'.

Google described its version in 2012 as understanding real-world entities and their relationships: 'things, not strings' 1. A string is just letters. A thing is the actual monument, person or business those letters refer to.

Its use of ambiguity shows the point. A search for 'Taj Mahal' could mean the monument or a musician, and plain keyword matching cannot tell them apart. A graph that knows both entities can ask which one you mean 1.

Explain it at your level

Think of a huge mind map where every circle is a real thing, like a person, a shop or a city, and every line says how two things are connected. A search engine reads that map to know that 'Taj Mahal' the building is different from 'Taj Mahal' the musician. When you search a famous name and see a box of facts on the side, that box is drawn from the map.

What did AI tools search for when we asked about it?

On 20 October 2026 we asked Gemini (with Google Search on) and a Claude agent with a web search tool. Gemini ran two searches: 'what is a knowledge graph' and 'how does Google use its knowledge graph'. The Claude agent ran six, which it had been told to keep between 6 and 10.

The Claude searches show the follow-up needs behind this question: how it works, how to get into a Google knowledge panel, how schema markup connects to entities, enterprise examples, and how a knowledge graph compares to entity SEO and Wikidata. The sources attached ranged from IBM and Wikipedia to Stanford, a graph-database vendor and several SEO software and agency sites.

Those follow-ups are the outline of this page. One engine and one run each, so treat the list as a sample.

How does Google use its Knowledge Graph?

Google uses its Knowledge Graph to understand what a search is about and to show summary information. When it launched on 16 May 2012 it held more than 500 million objects and more than 3.5 billion facts about them and their relationships 1.

By 20 May 2020 Google said it had amassed over 500 billion facts about five billion entities 2. It says the graph draws on hundreds of web sources plus open-source and licensed databases, and that Wikipedia is frequently cited but only one input 2.

The most visible output is the knowledge panel, the information box that appears beside or among results with a summary, key facts, photos and official or social links 2. Google says it aims to show the most relevant and popular information 2.

500 billionfacts about five billion entities, per Google in May 20202

How does a business or person get into the Knowledge Graph?

You cannot apply. Google's help page says its current policy does not manually create or delete knowledge panels, and that it cannot add a missing image by request 3. A panel appears when Google's systems have enough reliable information about you.

What you can do is be easy to recognise. Add Organization markup on your home page. Google says it helps Google understand your organisation's administrative details and disambiguate it, and it recommends name, url, logo and sameAs links to your other official profiles 5.

If a panel already exists, claim it. Google says many panels can be claimed by their subject through verification, after which you can suggest changes, and some corrections, such as social profile links, can be made directly 23. Local businesses use a separate Google business listing process (Google Business Profile, formerly Google My Business).

  • Use one exact business name, address and description across your site and profiles.
  • Publish an About page that states who you are, where you operate and who runs the company.
  • Add Organization markup with sameAs links to real profiles.
  • Earn independent coverage, such as news, directories and industry bodies.
  • Search your name in quotes monthly and check whether a panel appears.

What is Wikidata and is it the same as the Knowledge Graph?

They are different things. Wikidata is a free, collaborative, multilingual knowledge base run by the Wikimedia Foundation. Items have a Q-number, properties have a P-number, and a statement pairs a property with a value 4. Its example: Douglas Adams is item Q42, with an 'educated at' statement pointing to St John's College 4.

Google's graph is private and built from many inputs. Wikidata is public and openly licensed, so anyone can read it 4. It is one of many places a system might learn facts.

Not everyone belongs there. An item is accepted if it has a valid sitelink to a Wikimedia page, refers to a clearly identifiable entity that serious public references can describe, or fulfils a structural need 8. A new local firm with no independent coverage may not qualify, and you should not force it.

What does a knowledge graph look like for a small business?

A graph is a list of simple statements, each with a subject, a relationship and an object. Real examples include Google's own Knowledge Graph and Wikidata, where Douglas Adams is item Q42 4. Here is an invented example for a bakery. Its facts are made up for illustration.

Each line is one fact. A system can follow the links: from the bakery to its city, from its owner to her book, from the bakery to what it sells.

Illustrative graph statements (not a real business)
  • Greenfield Bakery --is a--> Bakery
  • Greenfield Bakery --located in--> Leeds
  • Greenfield Bakery --founded by--> Maya Patel
  • Maya Patel --author of--> 'Sourdough at Home'
  • Greenfield Bakery --sells--> Sourdough loaves
  • Greenfield Bakery --official profile--> https://example.com/greenfield

What do people get wrong about knowledge graphs?

The first myth is that schema markup puts you in the graph. Markup gives Google information, and Google says content owners can share some details through special markup 2. It does not force a panel.

The second myth is that you can request a panel. Google says it does not create panels on request 3. Wikidata also expects serious public references, so promotional entries are unlikely to last 8.

The third myth is that one graph rules everything. There is Google's, Bing's, Wikidata and many private ones inside companies. Work on being clear everywhere.

What an AI assistant searches when it answers this

We gave the question below to AI search tools and recorded the web searches they ran behind the scenes. This is called query fan-out. Each phrase is a page you could publish to be found and cited.

The prompt

“what is a knowledge graph and how does Google use it”

The searches it ran
  • knowledge graph definition explained
  • Google Knowledge Graph how it works
  • how to get into Google knowledge panel
  • schema markup entities knowledge graph SEO
  • knowledge graph examples enterprise
  • Knowledge Graph vs entity SEO Wikidata
  • what is a knowledge graph
  • how does Google use its knowledge graph

Observed October 20, 2026 using Gemini with Google Search and Claude web search. Claude agents were told to run 6 to 10 searches, so that count is our setting, not Claude's own choice. Assistants differ and change often, so treat this as a sample, not a rule. See how we collect it.

Frequently asked questions

What is the difference between a knowledge graph and a knowledge panel?

The knowledge graph is the underlying database of entities and facts. The knowledge panel is the box of information Google shows in search results, drawn from that data 2. A business can have graph data without having a panel.

Can I create my own Google knowledge panel?

No. Google's help page says its current policy does not manually create or delete knowledge panels 3. You can make your information clear and consistent, and verify and suggest changes if a panel already exists 23.

Does schema markup get me into the Knowledge Graph?

Not by itself. Organization markup helps Google understand and disambiguate your organisation, and Google recommends putting it on the home page 5. It is one clue among many, and no markup guarantees a panel.

How big is Google's Knowledge Graph?

In May 2020 Google said it had over 500 billion facts about five billion entities 2. At launch in 2012 it had more than 500 million objects and 3.5 billion facts 1. We found no newer official figure.

Do small businesses need Wikidata?

Not necessarily. Wikidata only accepts items that meet its notability rules, such as a Wikimedia sitelink or a clearly identifiable entity with serious public references 8. Focus first on your own site, markup and independent coverage.

About the author

Rahil Jain

Rahil Jain is an AI SEO (answer engine and generative engine optimisation) consultant and co-founder of Chalk Labs.

LinkedIn · Method and credentials

Sources

  1. Introducing the Knowledge Graph: things, not strings · Google (The Keyword blog), Amit SinghalOpened October 20, 2026. Supports: 'Things, not strings'; launch figures of more than 500 million objects and 3.5 billion facts; Taj Mahal ambiguity example. Posted 16 May 2012.
  2. A reintroduction to our Knowledge Graph and knowledge panels · Google (The Keyword blog), Danny SullivanOpened October 20, 2026. Supports: Over 500 billion facts about five billion entities; sources include hundreds of web sources and open and licensed databases; knowledge panel contents; many panels can be claimed by their subject through verification; markup can share some information. Posted 20 May 2020.
  3. Get verified on Google / knowledge panel help · Google Search HelpOpened October 20, 2026. Supports: Google's policy does not manually create or delete Knowledge Panels; verified users get feedback prioritised and some corrections made directly, such as social profile links.
  4. Wikidata:Introduction · Wikidata (Wikimedia Foundation)Opened October 20, 2026. Supports: Wikidata is a free, collaborative, multilingual, secondary knowledge base; items use Q-numbers, properties P-numbers; Douglas Adams Q42 example; data under CC0.
  5. Organization structured data · Google Search CentralOpened October 20, 2026. Supports: Organization markup helps Google understand administrative details and disambiguate the organisation; recommended properties include name, url, logo, sameAs; place on the home page. Last updated 2026-09-08.
  6. Web search tool · Anthropic (platform documentation)Opened October 20, 2026. Supports: Claude decides when to search, the API runs searches and returns results, and the answer carries cited sources. Documents web search; says nothing about Google's Knowledge Graph.
  7. Optimizing Your Content for Inclusion in AI Search Answers · Microsoft Advertising (Bing)Opened October 20, 2026. Supports: Schema is a type of code that helps search engines and AI systems understand your content. Published 8 October 2025.
  8. Wikidata:Notability · Wikidata (Wikimedia Foundation)Opened October 20, 2026. Supports: An item is acceptable if it has a valid sitelink to a Wikimedia page, refers to a clearly identifiable entity described by serious public references, or fulfils a structural need.
  9. General structured data guidelines · Google Search CentralOpened October 20, 2026. Supports: Markup must not describe content that is not visible to readers and JSON-LD must match the page body. Last updated 2026-07-10.
How this page was made. Built on Rahil Jain's AI SEO method and drafted with AI research assistance. Every statistic links to the page we read it on (opened October 20, 2026 or nearby). Published October 20, 2026; last updated October 20, 2026. Chalk Labs sells AI SEO services, so we have a commercial interest in this topic. Spotted an error? Tell us and we will fix it.